Ecological Economics 68 (2009) 1893–1898

Contents lists available at ScienceDirect

Ecological Economics

j ourna l homepage: www.e lsev ie r.com/ locate /eco lecon

ANALYSIS

Environmental amenities and optimal agricultural land use: The case of Israel

Iddo Kan a,c,⁎, David Haim b,c,1, Mickey Rapaport-Rom c,2, Mordechai Shechter c,d,3

a Department of Agricultural Economics and Management, The Robert H. Smith Faculty of Agriculture, Food and Environment, The Hebrew University of Jerusalem, P. O. B. 12,Rehovot 76100, Israelb Department of Agricultural and Resource Economics, Oregon State University, 1482 NW 20th Corvallis, OR 97330, USAc Natural Resource and Environmental Research Center, University of Haifa, Mt. Carmel, Haifa, 31905, Israeld Department of Economics, University of Haifa, Mt. Carmel, Haifa, 31905, Israel

⁎ Corresponding author. Tel.: +972 8 9489233/0; faxE-mail addresses: kan@agri.huji.ac.il (I. Kan), haimd

mickey.rapaport-rom@mail.com (M. Rapaport-Rom), sh(M. Shechter).

1 Tel.: +1 541 7520540.2 Tel.: +972 4 8240083; fax: +972 4 8240059.3 Tel.: +972 4 8249069; fax: +972 4 8240059.

0921-8009/$ – see front matter © 2009 Elsevier B.V. Adoi:10.1016/j.ecolecon.2009.01.006

a b s t r a c t

a r t i c l e i n f o

Article history:

This paper evaluates the effe Received 20 August 2008Received in revised form 12 January 2009Accepted 13 January 2009Available online 23 February 2009

JEL classification:Q10Q24Q50

Keywords:AgricultureLand-useAmenitiesPositive-mathematical-programming

ctiveness of changing land allocation among crops as a mechanism for increasingnet-social benefits, where production profits and amenity values are augmented. A positive mathematicalprogramming model is calibrated and applied to 43 regions in the northern part of Israel, using a crop-discriminating amenity-value function. Changes in land allocation increase net-social benefits by 2.4%nationwide and by up to 15% on the regional level, where in some regions the net-social-benefits-increase/profit–loss ratio exceeds 20. Therefore, the results indicate that a policy encouraging amenity-enhancementof agricultural land use is warranted, provided that it is implemented on a regional scale, rather than as acomprehensive nationwide-enforced program.

© 2009 Elsevier B.V. All rights reserved.

1. Introduction

The multi-functionality of farmland as a supplier of both agricul-tural products and other amenities, such as environmental habitatsand aesthetic landscapes, is well recognized, particularly in developedcountries. This is the driving force behind agricultural support policiesestablished to reward farmers for the external benefits they create(OECD, 2000, 2003; Peterson et al., 2002; EC, 2003), and thereby slowurbanization processes. The positive effects of agricultural amenitieshave been evaluated in various economic studies; examples includeHalstead (1984), Bergstrom et al. (1985), Beasley et al. (1986), Bowkerand Didychuk (1994), Hackl and Pruckner (1997), Ready et al. (1997),Fleischer and Tsur (2003), and Ready and Abdalla (2005). Theimplications of agricultural amenities in terms of land allocationbetween urban and rural uses have been studied byMcConnell (1989),Lopez et al. (1994) and Brunstad et al. (1999).

The levels of amenity services, however, are not uniform acrossagricultural land uses. Therefore, policies designed to encourage

: +972 8 9466267.@onid.orst.edu (D. Haim),echter@econ.haifa.ac.il

ll rights reserved.

agricultural preservation as a whole, without discriminating amongdifferent land uses, may result in suboptimal agricultural landallocation from society's point of view. Evidence for such differencesin the amenity benefits associated with diverse agricultural activitiesare provided by Drake (1992) and Brunstad et al. (1999); e.g., thelatter distinguish between tilled land, woodland and pasture. Morerecently, Fleischer and Tsur (2009) developed a unified framework forthe analysis of rural–urban land allocation that takes into account theheterogeneous amenity values of farmland across crops. Theyestimated separate demand functions for housing, productive agri-cultural land uses, and agricultural-based amenities. However, thedemand function for productive agricultural land was derived subjectto the assumption of a constant return to scale, under which theplanting area allocated to each crop is determined exogenously. Thisimplies that the share of land devoted to each crop is consideredconstant, unless it is urbanized. In other words, the analysis ignoresthe farmers' option to vary land allocations among crops in order tosubstitute the supply of agricultural products with additional amenityservices.

The objective of the present study is to investigate the potential forincreasing society's benefits through changing intra-agricultural landallocation among crops, while taking into account the variabilityamong crops with respect to both profitability and amenity contribu-tion. To this end, we develop a positive mathematical programming(PMP) model (Howitt, 1995) which, contrary to the constant-return-

1894 I. Kan et al. / Ecological Economics 68 (2009) 1893–1898

to-scale modeling approach, enables smooth variations in landallocation among crops.

The analysis is applied to Israel, utilizing the amenity-valuefunction estimated by Fleischer and Tsur (2009). This functionrepresents regional communities' willingness to pay for preservingvarious agricultural landscapes, owing to the amenity services theygenerate. The PMP model is used to assess the extent to which aregional policy encouraging extension of the agricultural area devotedto high-amenity crops is warranted, while taking into account theneed to compensate farmers for the associated profit losses.

A strict farmland-protection policy in Israel prevents the emer-gence of rural–urban land market equilibrium (Alterman, 1997;Feitelson, 1999), and real-estate development of rural areas is subjectto official authorization. For that reason, Israel constitutes a con-venient case study in the sense that changes in intra-agricultural landallocations can be analyzed separately from the rural–urban land-useallocation, which is assumed constant in the present study.

The evaluation focuses on the heavily populated northern half ofIsrael—the part above and to the left of the bold line depicted in Fig. 1.The spatial units of the analysis are the 43 regions within this analyzedarea, termed ‘natural zones’. Two practical considerations underlie theselection of this particular partitioning into regions. First, the IsraeliCentral Bureau of Statistics (ICBS) routinely publishes aggregatedregional data based on natural zones. Second, the amenity-valuefunction estimated by Fleischer and Tsur (2009) is based on thesespecific natural zones.

In Section 2 we describe the development of the PMP model andthe integration of farmers' profits and amenity values into a net-social-benefits4 objective function. Section 3 presents the results ofthe evaluation, and Section 4 concludes the paper.

2. The model

We employ a two-stage PMP approach, based on Howitt (1995,2005). In the first stage, the model is calibrated separately for eachnatural zone, such that it reproduces the land allocation observedthere, which is considered optimal under profit-maximization (PM)farming behavior. In the second stage, the objective function isreformulated such that it encompasses both the farmers' profits andthe amenity value to the region's population. Themodel then searchesfor the regional Pareto efficient5 (PE) agricultural land allocation andcalculates both the increase in net-social benefits and the production-profit loss relative to the PM solution.

2.1. Calibration

Our data encompass land allocation among 45 crops, which, inprinciple, can be grown in each of the 43 natural zones.6 Let lik(in hectares) be the land allocated to crop i, i=1, …, 45, in region k,k=1, …, 43. Accordingly, Lk≡(l1,k, …, l45,k) is a vector denoting the

land allocation in region k, and lk =P45i=1

lik is the total agricultural land.

Under the current situation in Israel, farmers are not rewarded for theamenities they provide. Hence, in the calibration stage of the PMPmethod, it is assumed that in each region k, the observedland allocation is an outcome of PM behavior. This land allocationalso represents the intra-agricultural market-equilibrium solution,

4 The term “net-social benefits” is used instead of “social welfare,” since improvementsare measured in monetary units.

5 Pareto efficiency refers to a situation under which potential Pareto improvementsaccording to the Kaldor–Hicks criterion are unachievable.

6 The calibrated programming approach was favored due to the difficulty of applyingland-use estimation methodologies (e.g., Wu and Segerson, 1995; Miller and Plantinga,1999) based on this level of disaggregation with respect to both regions and crops.

which is denoted Lkm(l1,km , …, l45,km ). Thus, Lkm constitutes a solution to

the problem:

maxl1;k ;:::;l45;k

Πk Lkð Þ =X45i=1

lik piyi − γik +12δiklik

� �� �s:t: lk V lk; ð1Þ

where Πk (in $/year) is the annual profit associated with (vegetative)agricultural activities in region k, lk̄ (in hectares) is the total agriculturalland constraint for region k, yi (in ton/hectare-year) denotes the(uniform) per-hectare annual yield, and pi (in $/ton) is the (uniform)output price of crop i. The term γik + 1

2 δiklik (in $/hectare-year)represents the per-hectare average production cost, which is expressedas a linear function of the crop's land area, lik. This function is used toindirectly reflect the impact of various unobserved factors considered byfarmers while contemplating their land allocation among crops,including the spatial variability of the soil quality, marketing andagronomic risks, managerial limitations, etc. Consequently, the totalproduction cost is described here as a quadratic function of lik, and,contrary to the constant-return-to-scale-based models, the PMP modelenables the optimal land allocation to be smoothly altered in response toexogenous shocks.

The only unknown parameters in Eq. (1) are γik and δik; they arecalibrated by employing the two-step calibration procedure devel-oped by Howitt (2005), which is described in Appendix A. Our datasetincludes regional land allocations among crops and nationwide pricesfor the year 2002, as reported by the ICBS (2002). Nationwide yieldswere obtained from the ICBS for the period 1992–2002, and used forcalculating the average and the maximumvariation of the per-hectareproduction for each crop. In addition, production costs were collectedfrom various reports published by the Israeli Ministry of Agricultureand Rural Development (2002) for the 45 crops under consideration.Production costs vary among regions due to differences in precipita-tion and surface-water constraints, as calculated by Rapaport-Rom(2006). All monetary values are in 2002 US dollars.

2.2. Net-social benefits and amenity value

The net-social benefits for each region,Wk (Lk), is formulated as anadditively separable function:

Wk Lkð Þ = Πk Lkð Þ + Ak Lkð Þ; ð2Þ

where Ak (Lk) (in $/year) denotes the value of benefits enjoyed by theregion's local population due to the amenities associated with theregion's agricultural land uses, as estimated by Fleischer and Tsur(2009). This amenity value is given by

Ak Lkð Þ = Nkak Lkð Þ; ð3Þwhere Nk is the number of households residing in the region and ak(Lk) (in $/household-year) is the annual amenity value for the region'srepresentative household. Note that the public-good nature of theamenities is reflected by the multiplication of the per-householdamenity value by the number of households residing in the region.

Fleischer and Tsur (2009) applied the double-bounded-dichot-omous-choice elicitation technique to a representative sample ofIsrael's urban population in order to estimate households' willingnessto pay for the preservation of vegetative agricultural land uses. In apreliminary study, based on three focus groups, they used photos ofover 30 crop landscapes for classifying the landscapes according totheir aesthetic values. This enabled testing the level of differentiationamong landscapes, and led to the conclusion that individuals candistinguish among the amenity values associated with three groups ofcrops: (1) orchards and citrus, (2) vegetables, field crops and non-agricultural natural open spaces, and (3) greenhouses. We adopt thisclassification and index each group, respectively, by n, n=1,2,3, anddenote by likn the land devoted to a crop assigned to group n. The total

regional land devoted to group n's crops is given by lkn =PIni=1

likn,

Fig. 1. Partitioning of Israel into Natural zones and regional populations.

1895I. Kan et al. / Ecological Economics 68 (2009) 1893–1898

where In denotes the number of crops in that group. Thus, while in ouranalysis farmers decide on land allocation among 45 crops, theamenity value is calculated based on the aggregated area devoted to

the three amenity-related crop groups (the areas of non-agriculturalnatural open spaces are incorporated into the calculation of theamenity value, but assumed constant). Fleischer and Tsur (2009)

Table 1Amenity-influential groups of crops and the estimated parameters of the amenity-valuefunction (Fleischer and Tsur, 2009)

Crop group Crops Parameters

Orchards and citrus Orange, grapefruit, lemon, apple,pear, peach, plum, table grape,wine grape, banana, olive(non-irrigated), olive (irrigated),almond, avocado, palm

ϕ11=2.2×10−2

ϕ21=7.5×10−7

ϕ31=2.1×10−4

θ1=−9.5×10−6

η12=−3.9×10−7

η13=1.3×10−5

Vegetables, field cropsand natural open spaces

Potato, tomato (open-field),eggplant, vegetable marrow,onion, carrot, lettuce, cabbage,cauliflower, celery, radish,artichoke, garlic, bean, wheat,barley, cotton, chickpea, corn,pea, groundnut, sunflower,winter forage, summer forage,natural open spaces

ϕ12=4.9×10−3

ϕ22=1.2×10−7

ϕ32=−8.0×10−5

θ2=−1.5×10−7

η23=2.5×10−6

Greenhouses Watermelon, sugar-melon,tomato (greenhouse), cucumber,cepper, strawberry

ϕ13=7.9×10−2

ϕ23=−1.7×10−6

ϕ33=−1.0×10−5

θ3=−2.2×10−4

Table 2Values of the net-social-benefits elements and land allocations under the PM and PEsolutions, aggregated over the 43 natural zones

Profit maximization(PM)

Pareto efficiency(PE)

Difference(PE)−(PM)

Net-social-benefits components (106 $/year)Production profits 456.5 454.0 −2.5Amenity value 212.4 231.0 18.6Net-social benefits 668.9 685.1 16.2

Land allocation (ha)Orchards and citrus 61,179 61,659 480

1896 I. Kan et al. / Ecological Economics 68 (2009) 1893–1898

estimated the following per-household quadratic amenity-valuefunction:

ak Lkð Þ =X3n=1

f n Xkð Þlkn + 1= 2θn lknð Þ2� �

+ η12lk1lk2 + η13lk1lk3

+ η23lk2lk3;

ð4Þ

where Xk is a vector of characteristics of region k's population. In thisspecification Xk=(x1k, x2k), where x1k, x2k are, respectively, theaverage income (in $/household-month) and the average age of theregion's heads of households, and where fn(Xk)=ϕ1n+ϕ2nx1k+ϕ3nx2k.The parameter θn, n=1,2,3, represents the own-quadratic effect, and isfound to be negative for all n. The cross-effect parameters η12, η13 andη23 can be of either sign, where a negative (positive) parameterrepresents substitution (complementary) relationships between thecorresponding crop groups. Table 1 presents the allocation of the 45crops among the three amenity-influential groups, and the parametersof the amenity-value function, estimated by Fleischer and Tsur (2009).Base-year agricultural land allocations, as well as the average head-of-household age and income, were obtained from the ICBS (2002). Thenon-agricultural natural open-space areas, which are assumed con-stant in the analysis, were taken from Frenkel (2001).

Inspecting the per-household amenity function, it appears that,due to the negativity of some of the estimated own- and cross-effectparameters, the marginal amenity value with respect to lkn,

Aak Lkð ÞAlkn

,n=1,2,3, diminishes with lkn. In other words, as also found by Lopezet al. (1994), the amenity-value function exhibits a decreasing returnto scale. This phenomenon points to the potential for scale effects, andhighlights the importance of applying the estimated function to thoseregions that formed the basis for estimating it in the first place, asdone here.

The programming model is built on an Excel worksheet and run bythe Premium Solver Platform V6.5 instrument. The program seeksthe global optimum by employing a multi-start search procedure. Itemploys a quasi-Newtonianmethod based on quadratic extrapolations,where central differencing is used to estimate partial derivatives.7

3. Application

The PE solution is computed by searching for the regionalagricultural land allocation among the 45 crops, Lk, which maximizes

7 The model and the entire dataset are available from the authors upon request.

the net-social-benefits function, subject to the regional land con-straint:

maxl1;k ;:::;l45;k

Wk Lkð Þ = Πk Lkð Þ + Ak Lkð Þ s:t: lk V lk: ð5Þ

We denote by Lks the PE land allocation. It is assumed that farmers

are willing to shift from the PM land allocation, Lkm, to Lks , as long as

they are at least fully compensated for the consequential profit loss. Afeasible policy for stimulating such a change may take the form of, forexample, compensation payments funded by the local authorities ineach region. Our analysis, however, is indifferent to the potentialimpact of diverse mechanisms of amenity-enhancing land policies(see Johnston and Duke, 2007) and focuses on an empirical evaluationof the farming profit loss and the increase in net-social benefitsassociated with the move from the PM to PE land allocation.

Table 2 presents the components of the net-social benefits and theland allocations under the PM and PE solutions, aggregated over the43 regions. Under the PM solution farmers' profits amount to$456.5 million/year, where the associated value of the amenities is$212.4 million/year. Thus, one-third of the net-social benefitsgenerated by the vegetative agricultural lands and natural openspaces throughout the northern part of Israel is attributed to environ-mental amenity services.

The agricultural sector's annual profit loss (nationally) incurred bymoving from the PM to the PE solution amounts to $2.5 million, wherethe annual increase associated with the amenity value is $18.6 million.In other words, on average, a 0.5% reduction in farming profits canenlarge the value of amenities by 8.5%. Put another way, for everydollar paid as compensation for farming profit losses, society wouldreap about $7.50 worth of benefits in the form of environmentallyenhanced agricultural amenities. Moreover, this increase in net-socialbenefits is achieved by a relatively small change in land allocation,which amounts to 1870 ha; i.e., land use is changed among the threeamenity-related groups of crops in only 0.5% of the total cultivatedarea under consideration—on average, the net-social-benefits increaseper hectare of land that switches uses is $1290. However, while thenet-social-benefits-increase/profit–loss ratio may seem relativelylarge, the overall change in net-social benefits amounts to only$16.2 million/year, which is an increase of merely 2.4% relative to thePM solution. This “return” may be viewed as too small to justify theenforcement of a nationwide agricultural landscape amenity enhance-ment program, particularly in view of the associated (unknown)implementation costs.

However, considering the issue at the regional level reveals widevariability among regions, with net-social-benefits increases varyingbetween 15% and no increase at all (Fig. 2).

Indeed, five regions alone, in which the rise exceeds $1 million ayear, account for almost 60% of the nationwide net-social-benefitsincrease, where in some regions the net-social-benefits-increase/profit–loss ratio goes beyond 20. Inspecting the objective function in

Vegetables, field cropsand open spaces

310,794 309,002 −1791

Greenhouses 16,189 17,579 1390

Fig. 2. Regional-scale profit reductions and increases in net-social benefits associated with the shift from the PM to the PE land allocation.

1897I. Kan et al. / Ecological Economics 68 (2009) 1893–1898

Eq. (5), this variability among regions is attributed to the integratedimpacts of three determinants: first, regions with larger differencebetween the PM land allocation and the allocation that maximizes theamenity value would face larger net-social-benefits enlargement;second, the impact of the amenity-value element grows with theregional population size; and third, the amenity value depends on theaverage income and age of the region's households. Compilation ofthese three factors determines the net-social-benefits increasespecific to each region. For example, the population in the naturalzone exhibiting the largest net-social-benefits rise—Petah Tiqwa(number 422), with over $3 million a year—is about half of thepopulation in the mostly populated zone (Judean Mountains,number 111), in which net-social benefits are enlarged by only$0.7 million a year—the smaller improvement in the latter isattributed to the relatively larger similarity between the PM and themaximum-amenity-value land allocations in that region.

The revealed considerable variability among regions providessupport for a national-scale regulation that would grant local

communities the authority to decide on whether to implementpolicies regarding the provision of agricultural amenities. Moreover,enabling flexibility in regional-scale policy making would allow localregulators to design amenity enhancing programs that fit to theirregions' specific characteristics.

4. Concluding remarks

Our analysis for the case of Israel indicates that a policy to enhanceagricultural amenities by encouraging changes in intra-agriculturalland allocations among crops has the potential to increase net-socialbenefits. This policy's potential, however, is contingent upon itsimplementation on a regional scale, rather than as a comprehensivenationwide-enforced program. This outcome signifies that in most ofthe analyzed regions, the preferred agricultural landscape from thepopulation's point of view is fairly similar to the observed landscape,i.e., to the assumed farming PM solution. This similarity may resultfrom the population's adaptation to its local surroundings—a possible

1898 I. Kan et al. / Ecological Economics 68 (2009) 1893–1898

topic for further research. On the other hand, the amenity-valuefunction adopted in this study represents only landscape contribu-tions to the local population. Hence, a potential objective of futurestudies is to assess the extent to which the incorporation of thebenefits to tourists and residents of neighboring regions can increasethe justifiability of agricultural amenity-enhancing policies.

Appendix A

The first calibration step is based on transforming the quadraticprogramming problem into a linear programming problem. This isdone by replacing the per-hectare average production-cost function,γik + 1

2 δiklik, with the parameter cik (in $/hectare-year), which is theobserved base-year average production cost. The resultant linear

objective function, πk =P45i=1

lik piyi − cik½ �, should be maximized

accordingly by setting the land allocation, Lk, subject to the regionaltotal land constraint, lk≤ lk̄, and an additional set of 45 auxiliary land-calibration constraints, lik≤ lik

m+ε, where likm is the base-year observed

(PM) land allocated to crop i, and ε is a perturbation element, the roleof which is to ensure the effectiveness of the total land constraint,lk≤ lk̄. This linear-programming stage yields the dual values of the 46constraints, which are used for calculating γik and δik in the secondcalibration step.

Let i=1 denote the crop with the lowest observed average per-hectare profit. Thus, λk1 (in $/hectare-year)—the dual value of theregional total land constraint lk≤ lk̄—is given by λk

1=p1(y1−Δy1)−c1k,where Δy1 is crop 1's maximum observed yield reduction below itsaverage yield, y1. Then, λik2 (in $/hectare-year), which stands for the dualvalue of the auxiliary land-calibration constraint with respect to cropi, lik≤ lik

m+ε, i=1, …, 45, can be calculated according to λik2=piyi−

cik−λk1. Using the observed Lk

m land allocation, and substituting theequality cik = γik + 1

2 δiklmik, we get δik=2λik2 /likm and γik= cik−λik

2 forevery crop i, i=1, …. 45.

References

Alterman, R., 1997. The challenge of farmland preservation: lessons from a six-nationcomparison. J. Am. Plan. Assoc. 63, 220–243.

Beasley, S., Workman,W.,Williams, N.,1986. Amenity values of urban fringe farmland: acontingent valuation approach. Growth Change 17, 70–78.

Bergstrom, J., Dillman, B., Stoll, J., 1985. Public environmental amenity benefits ofprivate land: the case of prime agricultural land. South. J. Agric. Econ. 17, 139–149.

Bowker, J., Didychuk, D., 1994. Estimation of nonmarket benefits of agricultural landretention in eastern Canada. Agric. Res. Econ. Rev. 23, 218–225.

Brunstad, R.J., Gaasland, I., Vårdal, E., 1999. Agricultural production and the optimallevel of landscape preservation. Land Econ. 75, 538–546.

Drake, L., 1992. The non-market value of the Swedish agricultural landscape. Eur. Rev.Agric. Econ. 19, 351–364.

EC Council Regulation no. 1782/2003. Online, October 2007: http://eur-lex.europa.eu/LexUriServ/site/en/consleg/2003/R/02003R1782-20060101-en.pdf.

Feitelson, E., 1999. Social norms, rationales and policies: reframing farmland protectionin Israel. J. Rural Stud. 15, 431–446.

Fleischer, A., Tsur, Y., 2003. Measuring the recreational value of open space. J. Agric.Econ. 54, 269–283.

Fleischer, A., Tsur, Y., 2009. The amenity value of agricultural landscape and rural–urbanland allocation. J. Agric. Econ. 60, 132–153.

Frenkel, A., 2001. Open versus Development: the Loss of Agricultural Land in Israel inthe 2000's. The Technion, Haifa (in Hebrew).

Hackl, F., Pruckner, G.J., 1997. Towards more efficient compensation programs fortourists' benefits from agriculture in Europe. Environ. Res. Econ. 10, 189–205.

Halstead, J.,1984.Measuring the nonmarket value ofMassachusetts farmland. J. Northeast.Agric. Econ. Counc. 13, 12–19.

Howitt, R.E., 1995. Positive mathematical programming. Am. J. Agric. Econ. 77, 329–342.Howitt, R.E., 2005. Agricultural and environmental policy models: calibration, estima-

tion and optimization. Online, November 2005: http://www.agecon.ucdavis.edu/aredepart/facultydocs/howitt/master.pdf.

ICBS, 2002. Statistical Abstract of Israel, vol. 53. The State of Israel, Jerusalem.(in Hebrew).

IsraeliMinistry of Agriculture and Rural Development, 2002. Instructions for Growers. TheAgricultural Extension Service, Israel. http://www.shaham.moag.gov.il/ (in Hebrew).

Johnston, R.J., Duke, J.M., 2007. Willingness to pay for agricultural land preservationand policy process attributes: does the method matter? Am. J. Agric. Econ. 89,1098–1115.

Lopez, R.A., Shah, F.A., Altobello, M.A., 1994. Amenity benefits and the optimal allocationof land. Land Econ. 70, 53–62.

McConnell, K.E., 1989. Optimal quantity of land in agriculture. Northeast. J. Agric. Res.Econ. 18, 63–72.

Miller, J.D., Plantinga, J.A., 1999. Modeling land use decisionwith aggregated data. Am. J.Agric. Econ. 81, 180–194.

OECD (Organization for Economic Co-operation and Development), 2000. Multi-functionality: Towards an Analytical Framework. OECD, Paris.

OECD, 2003. The Future of Rural Policy—From Sectoral to Place-Based Policies in RuralAreas. OECD, Paris.

Peterson, J.M., Boisvert, R.N., de Gorter, H., 2002. Environmental policies for amultifunctional agricultural sector in open economies. Eur. Rev. Agric. Econ. 29,423–443.

Rapaport-Rom, M., 2006. Economic assessment of climate-change impacts on thevegetative agriculture in Israel. Unpublished M.Sc. thesis, Department of NaturalResources and Environmental Management, The University of Haifa, Israel(in Hebrew, with English abstract).

Ready, R., Abdalla, C., 2005. The amenity and disamenity impacts of agriculture:estimates from a hedonic pricing model. Am. J. Agric. Econ. 87, 314–326.

Ready, R., Berger, M., Blomquist, G., 1997. Measuring amenity benefits from farmland:hedonic pricing vs. contingent valuation. Growth Change 28, 43–58.

Wu, J., Segerson, K., 1995. The impact of policies and land characteristics on potentialgroundwater pollution in Wisconsin. Am. J. Agric. Econ. 77, 1033–1047.